Thrombosis and haemorrhage in polycythaemia ... - Semantic Scholar

review

Thrombosis and haemorrhage in polycythaemia vera and essential thrombocythaemia M. A. Elliott and A. Tefferi 1 Division of Hematology, Mayo Clinic, Rochester, MN, USA

Summary Despite decades of clinical and laboratory research, relatively little has been accomplished concerning the pathogenesis as well as the identification of risk factors for thrombosis and bleeding in myeloproliferative disorders. In polycythaemia vera, the pro-thrombotic effect of an elevated haematocrit is well established. In contrast, thrombocytosis per se has not been similarly incriminated in essential thrombocythaemia. In both conditions, advanced age and the presence of a prior event identify thrombosis-prone patients. There is increasing evidence to suggest an additional role by leucocytes that might partly explain the antithrombotic effects of myelosuppressive therapy. A substantial minority of affected patients display reduced levels of high molecular weight von Willebrand protein in the plasma during extreme thrombocytosis and it is believed that this might explain the bleeding diathesis of such patients. Recent controlled studies support the therapeutic value of hydroxyurea and aspirin in essential thrombocythaemia and polycythaemia vera, respectively. The current communication will address the incidence, phenotype, pathogenesis, risk factors, prevention, and treatment of both thrombosis and haemorrhage in these disorders. Keywords: polycythaemia vera, essential thrombocythaemia, thrombosis, haemorrhage. Polycythaemia vera (PV) and essential thrombocythaemia (ET) are traditionally classified as myeloproliferative disorders (MPD), which is a broad category of clonal stem cell diseases that also includes myelofibrosis with myeloid metaplasia (MMM) and chronic myeloid leukaemia (Dameshek, 1951). Among the MPD, both ET and PV are favoured with relatively long median survivals (15+ years) as well as low transformation rates into either acute leukaemia or MMM (15-year risk of 10–15%) [Rozman et al, 1991; Gruppo Italiano Studio Policitemia (GISP) 1995; Brodmann et al, 2000; Tefferi et al, 2001; Passamonti, 2004]. However, clinical course in both instances

Correspondence: Dr Michelle Elliott, Division of Hematology, Mayo Clinic, 200 First Street SW, Rochester, MN 55905, USA.

is complicated by relatively frequent thrombohaemorrhagic episodes (Landolfi et al, 1995).

Incidence A precise estimate for the incidence of either thrombosis or haemorrhage in PV and ET has been difficult to establish (Schafer, 1984; Landolfi et al, 1995). Regardless, based on a general review of published material, the frequency of thrombosis (arterial, venous and microcirculatory) has been consistently greater than that of bleeding, as outlined in Tables I (manifestations at diagnosis) and II (manifestations reported during follow-up). At initial presentation, the reported incidence of thrombosis and bleeding in ET varied from 11–25% to 3Æ6–37% respectively. The corresponding figures for PV were 12–39% and 1Æ7–20%. Factors that could have accounted for the relatively wide range of values include patient selection, definitions of events, accuracy in data reporting, and the effect of therapy. More recent series tend to describe lower incidence figures that are probably a result of both a lead-time bias in diagnosis and improved therapy (Jensen et al, 2000a). Of note, the incidence of thrombohaemorrhagic events during followup (Table II) remain substantial and reflect continued inadequacies in current management. Among the large retrospective studies in ET, only one also evaluated a control population (Cortelazzo et al, 1990). The overall risk of thrombosis and bleeding in ET was 6Æ6%/patient-year and 0Æ33%/patient-year, respectively versus 1Æ2%/patient-year and zero in the control population.

Phenotype Thrombosis. Thrombosis in PV and ET occurs in arterial, venous or microcirculatory locations (Tables I and II). In general, arterial events predominate over venous events (Landolfi et al, 1995). Major arterial cerebrovascular or cardiovascular events and intra-abdominal (portal and hepatic) venous events are more common in PV while microcirculatory manifestations are more common in ET (Landolfi et al, 1995). Large vessel arterial events are the predominant cause of morbidity and mortality and include, in descending order of frequency, cerebrovascular accidents

E-mail: [email protected]

ª 2004 Blackwell Publishing Ltd, British Journal of Haematology, 128, 275–290

doi:10.1111/j.1365-2141.2004.05277.x

Review Table I. Thrombotic and haemorrhagic events in essential thrombocythaemia (ET) and polycythaemia vera (PV) reported at diagnosis.

n

Platelet · 109/l Asymptomatic Major Major arterial Major venous MVD Total bleeds (median/mean) (%) thrombosis (%) thrombosis* (%) thrombosis* (%) (%) (%) (major)

ET Bellucci et al (1986) 94 1200 Fenaux et al (1990) 147 1150 Cortelazzo et al (1990) 100 1135 Colombi et al (1991) 103 1200 Besses et al (1999) 148 898 Jensen et al (2000a) 96 1102 PV Berk et al (1981) 431 532 GISP (1995) 1213 NA Passamonti et al (2002) 163 357 Barbui and Finazzi (2003) 1638 398 (ECLAP) Najean et al (1987) 58 NA Passamonti et al (2003) 70 544

67 36 34 73 57 52

22 18 11 23Æ3 25 14

81 83 91 87Æ5 NA 85

19 17 9 12Æ5 NA 15

43 34 30 33 29 23

37 (3Æ2) 18 (4) 9 (3) 3Æ6 (1Æ9) 6Æ1 (NA) 9 (5Æ2)

NA NA 37 NA

13 34 34 38Æ6

100 66 64 75

0 33 36 25

NA NA 24 NA

20 (NA) NA 3 (NA) 8Æ1

21 47

12Æ1 24Æ3

28Æ6 70Æ6

71Æ4 29*Æ4

NA 15Æ7

1Æ7 4Æ3

MVD, microvascular disturbances; NA, not available. *Percentage of total major thrombotic events. Estimate according to GISP (1995).

Table II. Thrombotic and haemorrhagic events in essential thrombocythaemia (ET) and polycythaemia vera (PV) reported at follow-up.

n

Percentage of deaths Percentage of from thrombosis Major Major arterial Major venous MVD Total bleeds deaths from thrombosis (%) thrombosis (%)* thrombosis (%)* (%) (%) (major) haemorrhage (%) (%)

ET Bellucci et al (1986) 94 17 Fenaux et al (1990) 147 13Æ6 Cortelazzo et al (1990) 100 20 Colombi et al (1991) 103 10Æ6 Besses et al (1999) 148 22Æ3 Jensen et al (2000a) 96 16Æ6 PV Berk et al (1981) 431 25 GISP (1995) 1213 19 Passamonti et al (2002) 163 18Æ4 Barbui and Finazzi 1638 11Æ5 (2003) (ECLAP) Najean et al (1987) 58 10Æ3 Passamonti et al (2003) 70 11Æ4

62Æ5 86 71 91 94 69

37Æ5 14 29 9 6 31

17 4Æ1 NA 33 27Æ7 16Æ7

14 (3Æ2) NA (0Æ7) NA (1) 8Æ7 (5Æ8) 11Æ5 (4Æ1) 13Æ6 (7Æ3)

0 0 0 0 0 3Æ3

NA 62Æ5 80 70Æ4

NA 37Æ5 15 29Æ6

NA NA 10 NA

NA NA NA (1Æ8) NA

7 3Æ1 6 4Æ3

33Æ3 75

66Æ6 25

NA NA

1Æ7 NA

0 11

0 25 100 one pt (IAVT) 27Æ3 13Æ3 16Æ7 34 29Æ2 19 41 70 11

MVD, microvascular disturbances; IAVT, intraabdominal venous thrombosis. *Percentage of total major thrombotic events.

(stroke and transient ischaemic attacks), myocardial infarction, and peripheral arterial occlusion (Cortelazzo et al, 1990; GISP, 1995; Landolfi & Marchioli, 1997; Besses et al, 1999; Jensen et al, 2000a). Lower extremity deep venous thrombosis (DVT) and pulmonary embolism account for the majority of venous events in most, but not all, series (Najean et al, 1987). Of importance, patients with PV or ET have an unusually high rate of intra-abdominal (portal and hepatic) vein thrombosis 276

and together account for a substantial proportion of identifiable causes of these potentially catastrophic events (Anger et al, 1989a; Cardin et al, 1992; Denninger et al, 2000). In one large series, thrombosis of major abdominal vessels was seen in 14 of 140 (10%) patients with PV and three of 23 (13%) patients with ET (Anger et al, 1989b). Another study of 187 patients with ET reported that 60% of all venous events occurred in either an abdominal vein (n ¼ 10) or cerebral sinus (n ¼ 2) (Bazzan et al, 1999). Similarly, Lengfelder et al


Review (1998) reported portal/hepatic/splenic venous thrombosis in 10 of 143 (7%) consecutive ET patients, accounting for 55% of all observed venous thrombosis. A recent study investigating pro-thrombotic disorders among patients presenting with otherwise unexplained (i.e. without cirrhosis or hepatobiliary carcinoma) portal (n ¼ 36) or hepatic (n ¼ 32) vein thrombosis demonstrated an underlying MPD in 31 and 53% of the cases, respectively (Denninger et al, 2000). Interestingly, young patients appear to be particularly vulnerable to this complication (Najean et al, 1987; Perea et al, 2001). In one study, among 58 PV patients aged less than 40 years, five developed hepatic vein thrombosis (three at diagnosis and two during follow-up), which proved fatal in four (Najean et al, 1987; Perea et al, 2001). This observation was supported by a more recent study of similarly young patients with 29Æ2% prevalence of the specific complication (Perea et al, 2001). Proposed aetiologies for the particular anatomic localization include increased portal blood flow, congestive splenomegaly, and hepatic extramedullary haematopoiesis, although the pathogenesis remains to be defined (Schafer, 1984). The above set of observations warrant a careful evaluation for the presence or future development of a MPD in any patient that presents with abdominal vein thrombosis (Valla et al, 1985). In their study, Valla et al (1985) assessed the prevalence of overt and latent MPD in 20 patients with idiopathic hepatic vein thrombosis and demonstrated endogenous erythroid colony formation (i.e. a marker for clonal myeloproliferation) in 16 patients, including 13 women aged 18–45 years. Among these 16 patients, the conventional criteria for the diagnosis of either PV or ET were met in only two (Valla et al, 1985). It should be appreciated that endogenous erythroid colony formation alone is not sufficient to make a diagnosis of a chronic MPD (CMPD), and that the phenomena of growth factor independence as well as hypersensitivity may be a non-specific, intrinsic cell property that is generic to many clonal processes (Juvonen et al, 1987). Microcirculatory disturbances. Microcirculatory symptoms (headache, paraesthesia, erythromelalgia) are more common in ET than PV. Erythromelalgia is a microvascular thrombotic syndrome, which presents with unilateral or bilateral asymmetric erythema, congestion, and burning pain of the hands and feet. In the absence of therapy this may progress to acrocyanotic ischaemia and, eventually, frank peripheral gangrene (Michiels, 1997). Histopathological studies demonstrate platelet-rich arteriolar microthrombi with endothelial inflammation and intimal proliferation (Michiels et al, 1985; van Genderen et al, 1996a). Increased platelet consumption is demonstrated during attacks, as determined by studies of platelet turnover, and the thrombi contain abundant von Willebrand factor (VWF) in association with platelets, but minimal fibrin (van Genderen et al, 1995, 1996a). A characteristic feature is the prompt symptomatic resolution with the initiation of aspirin (Michiels et al, 1985). However,

not all patients respond to aspirin and some may require platelet cytoreduction in order to obtain relief (Regev et al, 1997). Other patients may experience non-localizing transient neurologic and visual disturbances which have not been as well characterized as erythromelalgia, but may also be the result of similar microvascular occlusion, as suggested by responsiveness to aspirin therapy and platelet cytoreduction (Michiels et al, 1993). Bleeding. Bleeding manifestations in both PV and ET involve primarily the skin and mucous membranes, suggesting defective primary haemostasis, and include ecchymosis, epistaxis, menorrhagia and gingival haemorrhage (Randi et al, 1991). Gastrointestinal haemorrhage occurs less frequently but can be severe, necessitating hospitalization and blood transfusion, and is often associated with the use of aspirin (Kessler et al, 1982; Colombi et al, 1991). This type of bleeding pattern is consistent with platelet defects (quantitative or qualitative) or von Willebrand Disease (VWD). In view of the clonal derivation of the platelet in MPD, a dysfunctional platelet has long been assumed to be the main culprit in the haemorrhagic manifestations. However, increasing data indicate that this may only indirectly be the case, as thrombocytosis may cause an acquired von Willebrand Syndrome (AVWS), that may account for a significant proportion of reported bleeding episodes (Budde et al, 1993; Budde & van Genderen, 1997). Furthermore, the general lack of correlation between platelet function abnormalities and clinical bleeding suggest alternate mechanisms might be involved (Schafer, 1984). Although reported, intra-articular, retroperitoneal, and deep intramuscular haematomas, such as seen in haemophilia, are distinctly unusual (Fey, 1985; Fenaux et al, 1990; Ishihara et al, 2000). Some haemorrhagic episodes may be directly or indirectly related to concomitant thrombotic complications. For example, bleeding from gastric or esophageal varices usually results from abdominal vein thrombosis-associated portal hypertension (Menon et al, 2004). A more common, as well as difficult to manage, scenario is bleeding secondary to the clinicallyindicated use of anticoagulants or antiplatelet agents (Kessler et al, 1982; Tartaglia et al, 1986).

Pathogenesis Erythrocytosis. The relationship between thrombosis, haematocrit, and in vitro parameters of blood viscosity and tissue perfusion is complex. The haematocrit has been shown to be the major determinant of whole blood viscosity in vitro, however, in vivo flow dynamics in blood vessels and arterial oxygenation play an important role (Pearson, 1997). In vivo, increased haematocrit level is associated with decreased cerebral blood flow rate (Kassum & Thomas, 1977). This is not only because of the change in viscosity seen at higher haematocrit levels but also the increase in arterial oxygen content that adjusts cerebral blood flow rates accordingly


277

Review (Barton et al, 1985). In contrast to PV, cerebral blood flow in hypoxic pulmonary disease complicated by an erythrocytosis is not so markedly reduced and this is likely to reflect cerebral vessel dilatation consequent upon hypercapnia (Wade et al, 1981). Under in vivo flow conditions, axial migration of red cells occurs with displacement of platelets to the mural plasmatic zone, exposing them to maximal vessel wall shearing forces (Beck & Eckstein, 1980). With increased haematocrit, the width of the plasmatic zone becomes narrower, allowing greater platelet-endothelial cell as well as platelet–platelet interactions (Turitto & Weiss, 1983). This effect is more marked at high shear rates, which is comparable with flow in arterioles and capillaries, than it is at low shear rates as seen in the venous circulation. At high shear rates, platelet binding sites are modified leading to enhanced VWF binding to the platelet glycoprotein Ib (Gp Ib) receptor and then with platelet activation, Gp IIb/IIIa receptors. Furthermore, at increasing platelet counts, the increased platelet concentration in the plasmatic zone at raised haematocrit values, might enhance platelet–platelet interactions (Huang & Hellums, 1993). Among patients with underlying vascular disease these enhanced interactions may contribute to the pathogenesis of thrombosis seen at a high haematocrit. Thrombocytosis. The platelet count per se has not been significantly correlated with thrombosis risk in either PV or ET (Buss et al, 1985; Bellucci et al, 1986; Wehmeier et al, 1991a; Lengfelder et al, 1998; Besses et al, 1999). However, current information suggests that, in high risk patients, lowering the platelet count to below 400 · 109/l might reduce the incidence of thrombotic events (Regev et al, 1997; Storen & Tefferi, 2001). However, it is not clear whether this is attributed to the platelet reduction per se or the overall effect of myelosuppression (Cortelazzo et al, 1995). Paradoxically, retrospective studies have suggested an association between bleeding diathesis and extreme thrombocytosis that might be partly explained by the occurrence of AVWS in some patients (Buss et al, 1985; Bellucci et al, 1986; Fenaux et al, 1990; van Genderen & Michiels, 1994). Functional and structural platelet abnormalities. In routine clinical practice, platelet function is most commonly assessed by means of platelet aggregation studies and these have been explored over the last several decades in an attempt to better understand the pathogenesis of bleeding and thrombosis in the CMPD. An elusive goal has been to identify patients at risk for haemostatic complications, so that the appropriate therapy may be instituted prior to the occurrence of a devastating vascular event (Schafer, 1984; Wehmeier et al, 1990; Balduini et al, 1991). Unfortunately, although platelet aggregation studies are frequently abnormal (demonstrating either or both hypo- and hyperfunction), a disappointing lack of clinical correlation with haemostatic (either bleeding or thrombosis) complications has been the rule (Barbui et al, 1983; Fenaux 278

et al, 1990; Wehmeier et al, 1990). Several reasons may account for this, including the potential for ex-vivo activation during platelet-rich plasma preparation, observation of co-existent platelet hypo and hyperreactivity in the same patient and changes in these platelet aggregation patterns in individuals during the course of the disease (Baker & Manoharan, 1988; Remaley et al, 1989; Balduini et al, 1991). The latter phenomenon implies that observations made at one time point may not be expected to correlate with previous, or predict future, haemostatic events (Baker & Manoharan, 1988). The most commonly reported abnormalities include decreased primary and secondary aggegation patterns to either or both epinephrine and ADP and decreased response to collagen, in descending order of frequency, with generally normal responses to arachidonic acid (Bellucci et al, 1986). In a review of several studies in CMPD patients (not on antiplatelet medications) decreased aggregation responses to one, two or all of these agonists were noted in 57, 39 and 37%, respectively (Schafer, 1984). Spontaneous platelet aggregation is another characteristic finding, and initially reported as a possible predictor of thrombotic risk (Wu, 1978). Here again, multiple additional studies have failed to find any predictive value (Barbui et al, 1983; Fenaux et al, 1990; Wehmeier et al, 1990). Acquired storage pool deficiency is a characteristic feature of MPD and, unlike the congenital form (Weiss et al, 1979), it is probably a manifestation of abnormal in-vivo platelet activation (vide infra) with resultant granule release (Wehmeier et al, 1990). Evidence for platelet activation in MPD comes from demonstration of increased levels of plasma and urinary arachidonate metabolites [thromboxane B2 (TX B2)], a granule proteins (platelet-derived growth factor, b-thromboglobulin, platelet factor 4), and membrane markers of platelet activation (Burstein et al, 1984; Gersuk et al, 1989; Wehmeier et al, 1989, 1991b). The latter is usually assessed by flow cytometry using monoclonal antibodies to p-selectin, thrombospondin, and the activated fibrinogen receptor, GPII b/ IIIa) (Jensen et al, 2000b). The expression of activation-dependent platelet membrane proteins was assessed in samples from 50 MPD patients and 30 controls (Landolfi et al, 1995; Jensen et al, 2000b). Compared with controls, the mean percentage of p-selectin-positive and thrombospondin-positive platelets (both alpha-granule constituents) was significantly increased on platelets from MPD patients, and even more in patients who had experienced a thrombotic event. MPD-specific defects in arachidonic acid metabolism might result in abnormal and sustained thromboxane A2 (TX A2) generation (Schafer, 1982; Landolfi et al, 1992; Rocca et al, 1995). Cyclooxygenase-1 (COX-1) converts arachidonic acid to TX A2, which is potent vasoconstrictor and inducer of platelet aggregation, and is effectively inhibited by low dose aspirin that also alleviates microvascular symptoms (Michiels et al, 1985, 1996; Landolfi et al, 1992; Rocca et al, 1995). Although endogenous TX A2 had not been clinically proven to predict thrombotic complications, the recent demonstration,


Review in a randomized setting, of the value of low dose aspirin in reducing thrombosis risk in PV lends support for the clinical significance of spontaneous TX A2 generation in the pathogenesis of the thrombotic phenotype observed in PV and ET (Landolfi et al, 2004). Currently, the pathogenesis of the enhanced TX A2 production in MPD is unknown. Earlier work by Schafer (1982) demonstrated that a large proportion (40%) of patients with MPD had deficiency of lipoxygenase, an enzyme mediating the alternate pathway to cyclooxygenase metabolism of platelet arachidonate. Among patients with lipoxygenase deficiency, the majority demonstrated increased synthesis of thromboxane, hypothesized to be related to preferential shunting of arachidonate to the cyclooxygenase pathway (Schafer, 1982). Intuitively, this would be expected to result in a pro-thrombotic phenotype. However, the patients with lipoxygenase deficiency demonstrated a predominately haemorrhagic phenotype and also had a lower incidence of thrombotic events (Schafer, 1982). Platelet membrane receptor abnormalities. Multiple platelet surface membrane protein and receptor abnormalities have been demonstrated in MPD. One of the earliest observations was of decreased adrenergic receptor expression, felt to be the culprit for the frequently observed decreased or absent response to epinephrine in aggregation studies (Kaywin et al, 1978). Additional studies indicated that the latter might, in some cases, be because of impaired receptor signal transduction (Ushikubi et al, 1990). Platelet membrane GP Ib and GP IIb/IIIa receptors have been shown to be decreased in PV and ET using quantitative radiolabelled and, more recently, flow cytometric fluorescent, monoclonal antibody techniques (Mazzucato et al, 1989; Le Blanc et al, 1998; Jensen et al, 2000b). In spite of in vitro demonstration of deficiency of these key receptors of effective haemostasis, no correlation with clinical bleeding has been demonstrated and these abnormalities have not been shown to have any clinical role in identifying patients at risk of bleeding. In contrast, surface expression of GPIV is increased, and claimed to correlate with a history of prior thrombosis (Jensen et al, 2000b). Leucocyte activation. As disorders of the haematopoietic stem cell, the potential role of clonal leucocytes in the pathogenesis of thrombosis in PV and ET has received recent attention and would be consistent with the wellestablished anti-thrombotic effect of myelosuppressive therapy (Berk et al, 1981; Cortelazzo et al, 1995). A series of neutrophil activation parameters (CD11b, leucocyte alkaline phosphatase, cellular elastase, plasma elastase, and myeloperoxidase) as well as markers of both endothelial damage [thrombomodulin, VWF antigen (VWF:Ag)] and thrombophilic state (thrombin-antithrombin complex, prothrombin fragment 1 + 2, D-dimer) were studied and the results suggested the frequent occurrence of neutrophil

activation, as well as high plasma levels of endothelial and hypercoagulation markers in both PV and ET, compared with controls (Falanga et al, 2000). Another study using whole blood flow cytometry determined the presence of circulating platelet-leucocyte aggregates in 50 patients with MPD and 30 healthy controls (Jensen et al, 2001). Compared with controls, the mean percentage of platelet-leucocyte aggregates was increased in MPD patients and this finding was correlated to platelet count, percentage of P-selectin and thrombospondin-positive platelets and platelet expression of GPIV. Among the MPD patients, a history of either microvascular disturbances or a thrombotic event was associated with an even higher mean percentage of plateletleucocyte aggregates (Jensen et al, 2001). Although preliminary, the above set of observations offer an alternative and/or contributing mechanism of thrombosis in MPD. Acquired von Willebrand syndrome. The clinical association between MPD and AVWS was derived from several case series describing abnormalities of VWF in patients with extreme thrombocytosis and exhibiting mucocutaneous haemorrhagic manifestations similar to patients with congenital VWD (Budde et al, 1984; van Genderen et al, 1994; Michiels et al, 2001). Subsequently, several investigators have established a relationship between the platelet count and loss of large VWF multimers in plasma (Sato, 1988; Budde et al, 1993; Budde & van Genderen, 1997). VWF is a large multimeric glycoprotein that plays a key role in haemostasis (Ruggeri, 2001). It mediates the initial tethering and adhesion of platelets to sites of vascular injury, as well as subsequent platelet aggregation. The 225 kDa native subunits of VWF undergo polymerization into unusually large multimers within endothelial cells and megakaryocytes, and is stored in Weibel Palade bodies and alpha-granules at each of these sites, respectively (Ruggeri, 2001). Regulation of VWF size and function (large multimeric forms are most haemostatically effective) is based on proteolytic modifications, which occur soon after secretion or release into the bloodstream. This is mediated by a recently identified specific VWF cleaving protease (ADAMTS13) (Furlan et al, 1996; Tsai, 1996; Levy et al, 2001). In vivo the cleavage site for ADAMTS13 in the intact VWF subunit is normally protected but becomes susceptible to proteolysis under conditions affecting the VWF protein confirmation, such as high shear in the microcirculation. Deficiency or dysfunction of VWF results in a bleeding disorder known as VWD, which is usually congenital but may be acquired (Sadler et al, 2000). The AVWS described in MPD patients is characterized by the loss of large VWF multimers, which results in a functionally more relevant defect that may not be apparent when measuring VWF: Ag and FVIII levels alone, as these might remain within the normal range (Budde et al, 1984; Michiels et al, 2001). On the other hand, the assays used to assess VWF function, including collagen binding activity (VWF:CBA) and ristocetin cofactor activity (VWF:RCoA), show that VWF


279

Review function declines with increasing platelet counts (van Genderen et al, 1996b, 1997b; Favaloro, 2000). This phenomenon is not restricted to MPD and has also been described in patients with reactive thrombocytosis, suggesting a primary effect of absolute platelet number as opposed to dysfunctional clonal platelets (Budde et al, 1993). This would be consistent with the observed resolution of the specific laboratory abnormality with platelet cytoreduction (Budde et al, 1984; Budde et al, 1993; van Genderen et al, 1997b; Michiels et al, 2001). The exact mechanism of AVWS in MPD is unknown. The possibility of defective VWF synthesis or release is negated by studies that have demonstrated an appropriate increase in VWF:Ag and function following the administration of desmopressin (Budde et al, 1984; Lopez-Fernandez et al, 1987; van Genderen et al, 1997c). Likewise, platelet VWF has a normal multimeric distribution in both primary and reactive thrombocytosis, implicating a process that occurs following release into the circulation (Budde et al, 1984; van Genderen et al, 1997b). The loss of the large multimers appears to be due to decreased survival following secretion (van Genderen et al, 1997b). Increased clearance of large multimers has been proposed by some investigators to be the result of enhanced binding to platelets. However, the binding of VWF to single platelets from patients with ET and patients with reactive thrombocytosis was not significantly different from that of normal controls, as determined by flow cytometry (van Genderen & Leenknegt, 1999). Several investigators have described enhanced proteolysis of VWF in ET and other states of thrombocytosis, as judged by an increase in the 140- and 176-kDa proteolytic fragments of VWF. These fragments are derived from proteolytic cleavage, mediated physiologically by the ADAMTS13 cleaving protease (Budde et al, 1984, 1986; Lopez-Fernandez et al, 1987; Tsai, 1996; Levy et al, 2001). Assuming that the platelets are involved in increased proteolysis of the largest VWF multimers, it remains to be defined how this occurs. In states of thrombocytosis the increase in platelet number may facilitate the interaction between platelet surface receptor GP Ib and VWF, thus inducing the appropriate conformational changes in VWF that allow ADAMTS 13 to gain access to its cleavage site. At higher platelet numbers, this interaction may be enhanced, inducing excessive physiologic cleavage. One study attempted to define the bleeding tendency in relation to thrombocytosis as well as VWF abnormalities in patients with ET and reactive thrombocytosis compared with normal controls by assessment of the effect of aspirin on the bleeding time (van Genderen et al, 1997c). Treatment with aspirin resulted in bleeding time prolongation in all subjects but was more excessive in ET. All the patients with ET and excessive prolongation of bleeding time were found to have significantly reduced levels of large VWF multimers. These observations suggest that detection of AVWS in MPD might identify patients that are prone to aspirin-induced bleeding diathesis (van Genderen et al, 1997a). 280

Risk factors Age Advanced age is a general risk factor for thrombosis under any circumstance (Rosendaal, 1997). One large epidemiologic study of 1213 PV patients reported an overall rate of thrombotic events of 1Æ8 of 100 patients per year for those younger than 40 years of age, increasing to 5Æ1 of 100 patients per year for those older than 70 years (GISP, 1995). The largest and most recent PV epidemiologic study (n ¼ 1638), reported a hazard ratio of vascular complications of 8Æ6 for patients older than 60 years than in younger patients (P < 0Æ0001) (Barbui & Finazzi, 2003). Similar data exist for ET, providing further evidence for similar pathogenesis of thrombosis. One retrospective controlled trial of 100 consecutive ET patients reported an incidence rate of thrombosis of 1Æ7% (per patient per year) for those younger than 40 years, 6Æ3% for those aged 40–60 years, and 15Æ1% for those older than 60 years with ET. The rate of thrombosis in the control group was 1Æ2% per patient per year (Cortelazzo et al, 1990). Another more recent series of 148 ET patients reported the probability of a major vascular complication at 6 years of 35Æ6% for those older, vs. 21Æ4% for those younger, than 60 years of age. Only one of 36 (2Æ8%) patients younger than 45 years developed an event during follow-up (Besses et al, 1999). History of thrombosis. In both PV and ET, a previous thrombotic event has consistently proven to be an independent and predictive factor in determining recurrent thrombosis. This, in combination with patient age, provides the current cornerstone for risk stratification and is an indication to begin cytoreductive therapy, regardless of other factors. In the follow-up of 1213 PV patients, the frequency of thrombosis was 17Æ3% for patients without vs. 26Æ5% with a history of thrombosis (GISP, 1995). In the recently completed European Collaboration on Low-dose Aspirin in Polycythaemia Vera (ECLAP) epidemiology study of PV patients, a previous thrombotic event increased the hazard ratio of recurrence by 4Æ85 (P ¼ 0Æ0099). The interaction of increasing age (>60 years) and a prior thrombosis predicted a hazard ratio of 17Æ3 (P < 0Æ0001) (Barbui & Finazzi, 2003). Similar data exist for ET (Cortelazzo et al, 1990; Colombi et al, 1991; Besses et al, 1999). In a large controlled retrospective study of ET patients, the risk of thrombosis was 3Æ4% per patient-per year in those without vs. 13Æ4% per patient-per year in those with a previous thrombotic event (Cortelazzo et al, 1990). Cardiovascular risk factors. The contribution of wellestablished risk factors of cardiovascular disease (hypertension, smoking, hypercholesterolaemia, diabetes mellitus) have been assessed in multiple studies with some conflicting results, possibly reflecting the size of the studies and number of patients with the risk markers of interest included. On multivariate (hypertension, smoking,


Review hypercholesterolaemia, diabetes) analysis, one large retrospective study of ET patients (n ¼ 148) demonstrated the independent contribution of only hypercholesterolaemia, while another study (n ¼ 132) showed the independent association with smoking (Besses et al, 1999; Jantunen et al, 2001). The latter finding had been shown in an earlier, smaller (n ¼ 46) study (Watson & Key, 1993). A recent cohort study of 69 female ET patients indicated hypertension to be significantly correlated with thrombotic events (Shih et al, 2002). However, other large retrospective studies that included these variables could not confirm these associations (Cortelazzo et al, 1990; Fenaux et al, 1990; van Genderen et al, 1997d; Jensen et al, 2000a). The recent ECLAP epidemiology study of 1636 PV patients found that, in addition to age and prior thrombotic events, significant predictors of survival and cardiovascular mortality were smoking habit, diabetes, and congestive heart failure (Barbui & Finazzi, 2003). Hereditary and acquired thrombophilic states. Over the last two decades there has been a vast expansion in the knowledge of hereditary and acquired conditions contributing to thrombosis (Van Cott et al, 2002). These include the congenital deficiencies of natural anticoagulants (antithrombin, Protein C and Protein S), genetic mutations [Factor V Leiden, Prothrombin G20210A, and methylenetetrahydrofolate reductase (MTHFR) mutations] and acquired conditions (anticardiolipin antibodies and/ or lupus anticoagulants). Several recent studies have explored the contribution of these to the occurrence of thrombotic events in MPD. Two prospective studies evaluated ET patients to determine the allele frequencies of factor V Leiden, prothrombin G20210A, and MTHFR mutations among those with and without thrombotic complications (Dicato et al, 1999; Afshar-Kharghan et al, 2001). No significant correlation was found in either study, although both were limited by sample size (n ¼ 43 and n ¼ 42 respectively). A more recent larger, yet retrospective study of 304 patients with PV (n ¼ 178) and ET (n ¼ 126), demonstrated a significant difference in the prevalence of the Factor V Leiden mutation in patients with (16%), compared with those without (3%), a history of venous thrombotic events (Ruggeri et al, 2002). Importantly, carrier state for the Factor V Leiden was significantly associated with a higher risk of recurrent venous thrombotic events during follow-up (prevalence of 3Æ6% in asymptomatic patients, 6Æ9% in patients with a single episode and 18Æ1% in patients with recurrent venous thrombosis, respectively) (Ruggeri et al, 2002). No such relationship was seen with regards to arterial thrombosis, as would be expected with this mutation (Press et al, 2002). This study did not investigate the other common thrombophilic polymorphisms (prothrombin G20210A, MTHFR) or acquired states (antiphospholipid syndrome), known to effect the phenotypic expression of pro-thrombotic disease (Ruggeri et al, 2002).

Several studies have demonstrated elevated homocysteine levels among patients with MPD (Gisslinger et al, 1999; Faurschou et al, 2000; Amitrano et al, 2003). Elevated homocysteine levels may be hereditary (homozygosity for MTHFR mutation) or acquired (folic acid, vitamin B12 or B6 deficiencies, renal failure), and is a recognized risk factor for both arterial and venous thrombotic disease (Key & McGlennen, 2002). However, no association with the homozygosity for the MTHFR mutation could be confirmed in these studies, leading investigators to postulate (and one study confirm) that the acquired increase in homocysteine may be the result of vitamin B deficiencies, possibly related to the persistent hyperproliferative haematopoiesis of these disorders. With regards to clinical impact of these observations, an association with arterial thrombotic disease was demonstrated in only one study (Amitrano et al, 2003) but the others could not validate any association with thrombotic disease (Gisslinger et al, 1999; Faurschou et al, 2000). Antiphospholipid antibodies, as measured by either by a clot-based assay (lupus anticoagulant) or enzyme-linked immunosorbent assay are established risk factors for both arterial and venous thrombotic disease (Galli et al, 2003). Several investigators have reported an increased prevalence of antiphospholipid antibodies in ET (Harrison et al, 2002; Jensen et al, 2002). One study suggested an association between the observed antiphospholipid antibodies and thrombosis (Harrison et al, 2002). However, the small sample sizes of patients included and lack of follow-up testing to confirm persistence (and thus significance) of the detected antibodies precludes any firm conclusions be made of these findings. Clonal haematopoiesis. X-linked clonal analysis has demonstrated that patients with ET could be classified into either a monoclonal or polyclonal pattern of X chromosome inactivation and that the former might be associated with an increased risk of thrombosis (el-Kassar et al, 1997; Harrison et al, 1999; Chiusolo et al, 2001; Shih et al, 2002). In the most recent, as well as largest of these studies, 69 patients with ET were investigated. The monoclonal group (68Æ4% of the study population), compared with the remaining patients (polyclonal group), were found to be at a greater risk of thrombosis (Odds ratio: 6Æ87) (Shih et al, 2002). However, there was a significant difference in the age distribution between the two groups and the observations can be considered valid only after a prospective evaluation.

Surgery and reduction of risks of thrombosis and bleeding Patients with CMPD have a higher risk of morbidity and mortality when undergoing surgical procedures. The risks are particularly high when the underlying myeloproliferation is poorly controlled as reflected by suboptimal control of erythrocytosis in PV or of thrombocytosis in both PV and ET. In view of the underlying pro-thrombotic phenotype of these disorders, the risks accompanying the perioperative


281

Review hypercoagulable state are magnified. Paradoxically, the risks of haemorrhage are also increased in these patients, particularly when thrombocytosis is poorly controlled. Postoperatively, a relative secondary thrombocytosis is seen in normal individuals following major surgery (Warkentin et al, 2003). This reactive phenomenon is likely to produce an even greater effect among patients with CMPD. Potentially, this may lead to the development or exacerbation of an AVWS and associated haemorrhagic diathesis. These risks are dramatically increased in the setting of splenectomy, a procedure often performed in these patients for management of symptomatic splenomegaly because of portal vein thrombosis or MMM (Malmaeus et al, 1986; Tefferi et al, 2000; Randi et al, 2002). This procedure carries a high risk of operative mortality (9%) and morbidity (31%) (Tefferi et al, 2000). In one large series, perioperative fatal and non-fatal bleeding occurred in 4Æ5 and 14Æ5% of MMM patients, respectively (Tefferi et al, 2000). Fatal and non-fatal major thrombotic events were reported in an additional 1Æ3 and 7Æ2%, respectively (Tefferi et al, 2000).

Management Most investigators agree that specific drug therapy does not cure the underlying disease or prevent clonal evolution in either PV or ET (Tefferi & Murphy, 2001; Tefferi, 2003; Barbui et al, 2004). Therefore, the current rationale for using drugs in these conditions is to either prevent or treat a thrombohaemorrhagic event. Prevention of thrombosis. In patients with PV, the risk of thrombosis correlates with the haematocrit, and phlebotomy to maintain this at